Grammarly's AI is recruiting writers, alive and dead, whether they like it or not

The latest unnecessary feature from an app that's supposed to just check for typos offers "Expert Reviews" from victims of identity theft. 

Grammarly's AI is recruiting writers, alive and dead, whether they like it or not

At least they’re not hiding it anymore. Facing accusations of plagiarizing human labor for slop, the AI industry has become much more brazen about what its models are trained on. ByteDance allegedly faked a fight between Brad Pitt and Tom Cruise to scare Hollywood screenwriters. Spotify is filled with AI knockoffs of artists protesting the company’s AI weaponry investments and ICE ads. Now Grammarly, which rebranded to Superhuman last year, fearing someone could possibly forget that it’s been enshitifying its own product with AI for the past few years, has launched a new feature that openly admits, “Yeah, we trained on real authors’ work without permission. Do something about it!” 

Grammarly’s latest complete waste of time, money, and resources isn’t a more advanced comma-splice detector, but rather a chance for AI Chatbots to learn the verbiage of writers, living or dead, under the guise of some throwaway advice. Last summer, Grammarly launched eight specialized AI agents to “provide targeted assistance for specific writing challenges.” The idea has since evolved to hoover up famous authors’ work. Earlier this week, Wired reported that using Grammarly’s “Expert Review” allows an approximation of Stephen King and Neil deGrasse Tyson to nitpick your work. While Tyson has the opportunity to say whether he’d like to be turned into a chatbot, other authors, like Carl Sagan, cannot because they are dead.

Today, The Verge reports that Grammarly’s Review has stolen the identities of its writers, as well as those of writers at The New York Times, The Atlantic, Gizmodo, and more. Unsurprisingly, it doesn’t do a very good job either. “The descriptions for some experts contain inaccuracies, such as outdated job titles,” writes Steven Bonfield, “which could have been accurately updated had Superhuman asked those people for permission to reference their work.” The disclaimer underneath Reviews warns, “References to experts in this product are for informational purposes only and do not indicate any affiliation with Grammarly or endorsement by those individuals or entities.” 

One person who did not indicate any affiliation with Grammarly is Verge Editor-in-Chief Nilay Patel, who wrote on Bluesky, “I am allllllmost more offended by the suggestion that I would give this shitbox edit than having my identity stolen.” Patel also wrote that Grammarly’s CEO is scheduled to appear on The Verge‘s Decoder podcast soon and “we will see if they back out.” He concludes, “Also it feels important to say there is literally no possible way to know what an editor is like as an editor by reading published written work by that person, which often goes through… other editors!”

 
Join the discussion...
Keep scrolling for more great stories.